Robust mesh tracking and fusion by using part-based key frames and priori model

1. A method comprising: segmenting an object in free viewpoint video content into parts, wherein segmenting is based on determining edges of the parts; tracking the parts of the object, wherein tracking the parts includes matching and storing position information for the parts, wherein tracking the parts of the object includes distinguishing key frames from non-key frames and matching a parametric model with a key frame or a non-key frame, wherein tracking the parts of the object includes generating keyframe pools, wherein each part is associated with a keyframe pool by identifying a pose of the part, wherein the keyframe pools utilize a hierarchical structure to connect the parts; stitching the parts of the object, wherein stitching the parts of the object includes stitching a first part from a first time and a second part from a second time, wherein the first time and the second time are different times, further wherein the second part from the second time is used when the first part from the first time is at least partially occluded, wherein the second part from the second time includes minimal or no artifacts as determined by locating artifacts and avoiding frames containing the artifacts; and integrating the parts of the object to regenerate the object.

2. The method of claim 1 further comprising identifying a failure of tracking of a part, including running a keyframe update wherein a locally re-meshed keyframe is extracted using a volumetric approach on a previous keyframe.

3. The method of claim 2 further comprising implementing part-based keyframe refinement, whole mesh tracking and re-meshing based on the parts of the object.

4. The method of claim 1 wherein segmenting the object into the parts includes segmenting the object into body parts.

5. The method of claim 1 wherein integrating the parts of the object ensures that the parts of the object are blended together to present an appearance of a single object.

6. The method of claim 1 further comprising acquiring the free viewpoint video content using at most 10 video cameras.

7. The method of claim 1 further comprising acquiring the free viewpoint video content using video cameras with at least 30 degrees of separation on every axis from other cameras.

8. An apparatus comprising: a non-transitory memory for storing an application, the application for: segmenting an object in free viewpoint video content into parts, wherein segmenting is based on determining edges of the parts; tracking the parts of the object, wherein tracking the parts includes matching and storing position information for the parts, wherein tracking the parts of the object includes distinguishing key frames from non-key frames and matching a parametric model with a key frame or a non-key frame, wherein tracking the parts of the object includes generating keyframe pools, wherein each part is associated with a keyframe pool by identifying a pose of the part, wherein the keyframe pools utilize a hierarchical structure to connect the parts; stitching the parts of the object, wherein stitching the parts of the object includes stitching a first part from a first time and a second part from a second time, wherein the first time and the second time are different times, further wherein the second part from the second time is used when the first part from the first time is at least partially occluded, wherein the second part from the second time includes minimal or no artifacts as determined by locating artifacts and avoiding frames containing the artifacts; and integrating the parts of the object to regenerate the object; and a processor coupled to the memory, the processor configured for processing the application.

9. The apparatus of claim 8 wherein the application is further for identifying a failure of tracking of a part, including running a keyframe update wherein a locally re-meshed keyframe is extracted using a volumetric approach on a previous keyframe.

10. The apparatus of claim 9 wherein the application is further for implementing part-based keyframe refinement, whole mesh tracking and re-meshing based on the parts of the object.

11. The apparatus of claim 8 wherein stitching the parts of the object includes stitching a first part from a first time and a second part from a second time, wherein the first time and the second time are able to be different times.

12. The apparatus of claim 8 wherein integrating the parts of the object ensures that the parts of the object are blended together to present an appearance of a single object.

13. The apparatus of claim 8 wherein the apparatus is further for acquiring the free viewpoint video content using at most 10 video cameras.

14. The apparatus of claim 8 wherein the apparatus is further for acquiring the free viewpoint video content using video cameras with at least 30 degrees of separation on every axis from other cameras.

15. A system comprising: a plurality of camera devices each configured for capturing free viewpoint video content; a first computing device configured for: segmenting an object in free viewpoint video content into parts, wherein segmenting is based on determining edges of the parts; tracking the parts of the object, wherein tracking the parts includes matching and storing position information for the parts, wherein tracking the parts of the object includes distinguishing key frames from non-key frames and matching a parametric model with a key frame or a non-key frame, wherein tracking the parts of the object includes generating keyframe pools, wherein each part is associated with a keyframe pool by identifying a pose of the part, wherein the keyframe pools utilize a hierarchical structure to connect the parts; stitching the parts of the object, wherein stitching the parts of the object includes stitching a first part from a first time and a second part from a second time, wherein the first time and the second time are different times, further wherein the second part from the second time is used when the first part from the first time is at least partially occluded, wherein the second part from the second time includes minimal or no artifacts as determined by locating artifacts and avoiding frames containing the artifacts; and integrating the parts of the object to regenerate the object; and a second computing device configured for: rendering the free viewpoint video content on the second computing device.

16. The system of claim 15 wherein the first computing device is configured for identifying a failure of tracking of a part, including running a keyframe update wherein a locally re-meshed keyframe is extracted using a volumetric approach on a previous keyframe.

17. The system of claim 16 wherein the first computing device is configured for implementing part-based keyframe refinement, whole mesh tracking and re-meshing based on the parts of the object.

18. The system of claim 15 wherein segmenting the object into the parts includes segmenting the object into body parts.

19. The system of claim 15 wherein integrating the parts of the object ensures that the parts of the object are blended together to present an appearance of a single object.

20. The system of claim 15 wherein the plurality of camera devices include at most 10 video cameras.

21. The system of claim 15 wherein the plurality of camera devices include video cameras with at least 30 degrees of separation on every axis from other cameras.